• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

whats going on with core 2 quad and windows?

Yeah then either the sensor is wonky or that card is overdrawing. Shouldn't exceed 50W at the slot IIRC.

That said, it may still be fine if the average stays under 50W (ie its only a short term "spike").

Well yes and no, hackers target modern operating systems due to the financial gains associated with contemporary security flaws and really have no interest in old hardware or software as those typically are 0 value targets

That also often aren't being patched actively. They are still useful for botnets too.

As long as your OS is still getting patches though don't worry about it.
 
Yeah then either the sensor is wonky or that card is overdrawing. Shouldn't exceed 50W at the slot IIRC.

That said, it may still be fine if the average stays under 50W (ie its only a short term "spike").



That also often aren't being patched actively. They are still useful for botnets too.

As long as your OS is still getting patches though don't worry about it. standard

PCIe standard is rated @75w this card doesn't have external power; the card is reporting 70.1 max sustaining 69w throughout the run its overclocked and undervolted as undervolting allows higher boost clocks and there is a significant boost in performance as the benchmark results indicate not worried as this machine lived a hard life playing Fortnite for years with no signs of degradation I'm tempted to test its limits in my AM 4 machine to see if its losing any performance if not then this theoretical 35w limit on dell's PCIe slot is debunked
 
That then. Mental memory is a pain lol.

You should be fine.
 
Is this power consumption accurate
No. The GTX 1050Ti is one of the nvidia cards where power sensing doesn't work properly. A give away is TDP (Total Default Power?) being 0% at load and nvidia-smi reporting N/A. TPU does pretty good reviews and uses external gear for power measurements so maybe you can get an idea from there.

Idle example
idle.png


Furmark
Furmark.png


Did you contact nvidia about popcnt problem? Maybe they would rectify it in newer releases if informed.
 
No. The GTX 1050Ti is one of the nvidia cards where power sensing doesn't work properly. A give away is TDP (Total Default Power?) being 0% at load and nvidia-smi reporting N/A. TPU does pretty good reviews and uses external gear for power measurements so maybe you can get an idea from there.

Idle example
View attachment 373077

Furmark
View attachment 373078

Did you contact nvidia about popcnt problem? Maybe they would rectify it in newer releases if informed.
The GTX 1050TI has a power rating of 75W as per TechPowerup.com. Using HWiNFO64, I've noted a maximum power draw of 70.1W and a current range of 65W-69W with a boost clock of 1759MHz during the Unigene Heaven benchmark, and 34W when idle. The source of those charts is unclear to me. Moreover, I've achieved these figures by undervolting to 925mV at 1759MHz. The original problem was fixed by rolling back to driver 552.44; drivers between 555.85 and 565.90 caused BSOD. The two most recent drivers operate correctly but fail to load Vulkan and OpenGL APIs, and I've used the Windows troubleshooting tool show hide updates to block Microsoft Nvidia driver updates.
 
The GTX 1050TI has a power rating of 75W as per TechPowerup.com. Using HWiNFO64, I've noted a maximum power draw of 70.1W and a current range of 65W-69W with a boost clock of 1759MHz during the Unigene Heaven benchmark, and 34W when idle.
34W idle is wrongly reported and is probably around 4W. Hopefully high power readings are somewhat closer.

The source of those charts is unclear to me.

The original problem was fixed by rolling back to driver 552.44; drivers between 555.85 and 565.90 caused BSOD.
Because the compiler used by nvidia for 555.85 to 565.90 introduced popcnt instructions which were not recognized by your CPU.

Nvidia fixed up the install by replacing popcnt instructions with a compatible routine in 566.03 and above however it seems they forgot to fix up "OpenGL" as "nvoglv64.dll" still contains popcnt, maybe some others too. The compiler replaces high level code instructions with machine code used by the processor. IOW nvidia software programmer didn't explicitly ask for popcnt to be used but instead the newer compiler threw them in there and seemingly nobody checked.

On a side note your GPUz screenshot shows PCIe socket voltage at 9V. Check what it is at idle. If 12V at idle that would indicate a power delivery problem at those loads.
 
On a side note your GPUz screenshot shows PCIe socket voltage at 9V. Check what it is at idle. If 12V at idle that would indicate a power delivery problem at those loads.
the 9v reading is at idle 1.6v at 100% usage stress test I ran time spy, and the score was verry good above average
 
I just downloaded DOOM played for about an Hr. HaHA, I have to say this old OptiPlex's performance is remarkable, achieving over 100 fps at 1080p with high detail settings using the Vulkan API. Both the CPU and GPU are operating at 90 to 100% capacity, yet there's no input lag, and the frame times are decent. The CPU and GPU seem to be well-matched, although the CPU temperature runs a bit hotter than I'd like, reaching the mid to high 70s°C, despite the TJ max rated up to 100°C . I'm thinking of sealing in-between fan shroud and the cooler housing since I reversed the fan, to avoid the CPU's hot air from venting into the case. I'm also considering installing a Noctua fan if I can find one that's comparable to the current Dell fan and running the proprietary sensor wire to the intake fan at the back of the case to circumvent a fan error at boot. Meanwhile, the GPU consistently reaches its maximum boost clock of 1759MHz undervolted at 925mv, and the GDDR5 memory is overclocked +730 on the slider in Afterburner, maintaining temperatures in the low 60s°C. I managed to upgrade the CPU cooler from an older Pentium4 socket 775 OptiPlex, as the stock 380 cooler was merely a chunk of aluminum; this upgraded one has a large copper core and heat pipe. I may also attempt to flat lap both the CPU and the copper contact on the cooler to improve conductivity, as they don't seem to be very flat.
Right, my bad

So misreading, wonder what HWiNFO shows. Nice score, is that the ASUS Phoenix card?
yes, it's the phoenix fan but with the shroud removed hardware info shows the same as afterburner and AIDA64
 
"Well, I'm so pleased with this OptiPlex. I just ordered some final touches. Also, I'm trying to find the Asus Cerberus backplate for this GTX 1050. If I can find one dirt cheap, this cooler has a larger copper plate as well as two heat pipes. This Noctua fan should not only quiet things down but flow better than the current Dell unit. The cable converts a 4-pin CPU fan to Dell's 5-pin proprietary specs. Though, I will also run a jumper wire from the 5th pin to the HDD caddy fan pin header to avoid the CPU fan error at boot."
 

Attachments

  • 710SA3TwpEL._AC_SX679_.jpg
    710SA3TwpEL._AC_SX679_.jpg
    35.4 KB · Views: 23
  • Screenshot (1).png
    Screenshot (1).png
    215.1 KB · Views: 33
  • Screenshot (2).png
    Screenshot (2).png
    362.4 KB · Views: 33
Last edited:
Back
Top